Lin, J. (1998). Factorizing multivariate function classes. In M. Kearns, M. Jordan, & S. Solla (Eds.), Advances in neural information processing systems, 10, (pp. 563-- 569). Cambridge, MA: MIT Press.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....as a (noisy) linear combination of sparse hidden variables s j , just as in the ICA model. However, the hidden variables are no longer completely independent, but can be divided into groups within which dependencies exist. Similar relaxations of the independence assumption in ICA were proposed in [20, 88]. Within these groups the dependencies take the form of correlations of energies: if one component s j in the group is significantly non zero, other components in that group are also likely to be active. This means that components in a group tend to be simultaneously active more often than would ....

J. K. Lin, "Factorizing multivariate function classes," in Advances in Neural Information Processing 10 (Proc. NIPS*97), MIT Press, 1998.

....as a (noisy) linear combination of sparse hidden variables s j , just as in the ICA model. However, the hidden variables are no longer completely independent, but can be divided into groups within which dependencies exist. Similar relaxations of the independence assumption in ICA were proposed in [20, 88]. Within these groups the dependencies take the form of correlations of energies: if one component s j in the group is significantly non zero, other components in that group are also likely to be active. This means that components in a group tend to be simultaneously active more often than would ....

J. K. Lin, "Factorizing multivariate function classes," in Advances in Neural Information Processing 10 (Proc. NIPS*97), MIT Press, 1998.

....that the s i can be divided into couples, triplets or in general n tuples, such that the s i inside a given n tuple could be dependent on each other, but dependencies between dioeerent n tuples were not allowed. Related relaxations of the independence assumption were proposed in (Cardoso, 1998; Lin, 1998). Inspired by Kohonen s principle of feature subspaces (Kohonen, 1996) the probability densities for the n tuples of s i were assumed in (Hyv#rinen and Hoyer, 2000) to be spherically symmetric, i.e. depend only on the norm. In other words, the probability density p q ( of the n tuple with ....

Lin, J. K. (1998). Factorizing multivariate function classes. In Advances in Neural Information Processing Systems, volume 10, pages 563569. The MIT Press.

....that the s i can be divided into couples, triplets or in general n tuples, such that the s i inside a given n tuple could be dependent on each other, but dependencies between dioeerent n tuples were not allowed. Related relaxations of the independence assumption were proposed in (Cardoso, 1998; Lin, 1998). Inspired by Kohonen s principle of feature subspaces (Kohonen, 1996) the probability densities for the n tuples of s i were assumed in (Hyv#rinen and Hoyer, 2000) to be spherically symmetric, i.e. depend only on the norm. In other words, the probability density p q ( of the n tuple with ....

Lin, J. K. (1998). Factorizing multivariate function classes. In Advances in Neural Information Processing Systems, volume 10, pages 563569. The MIT Press.

....it was assumed that the s i can be divided into couples, triplets or in general n tuples, such that the s i inside a given n tuple could be dependent on each other, but dependencies between dioeerent n tuples were not allowed. A related relaxation of the independence assumption was proposed in [2, 11]. Inspired by Kohonen s principle of feature subspaces [10] the probability densities for the n tuples of s i were assumed in [6] to be spherically symmetric, i.e. depend only on the norm. In fact, topographic ICA can be considered a generalization of the model of independent subspace analysis. ....

J. K. Lin. Factorizing multivariate function classes. In Advances in Neural Information Processing Systems, volume 10, pages 563569. The MIT Press, 1998.

....(x; y) of the feature subspace [11] In [11] it was shown that this principle, when combined with competitive learning techniques, can lead to emergence of invariant image features. Multidimensional independent component analysis. In multidimensional independent component analysis [2] see also [12]) a linear generative model as in Eq. 1) is assumed. In contrast to ordinary ICA, however, the components (responses) s i are not assumed to be all mutually independent. Instead, it is assumed that the s i can be divided into couples, triplets or in general m tuples, such that the s i inside a ....

J. K. Lin. Factorizing multivariate function classes. In Advances in Neural Information Processing Systems, volume 10, pages 563569. The MIT Press, 1998.

No context found.

References 15 Lin, J. K. 1998. Factorizing multivariate function classes. In Jordan, M. I., M. J. Kearns, and S. A. Solla, editors, Advances in Neural and Information Processing Systems, 10, pages 563--569, Cambridge, MA. MIT Press.

No context found.

Lin, J. (1998). Factorizing multivariate function classes. In M. Kearns, M. Jordan, & S. Solla (Eds.), Advances in neural information processing systems, 10, (pp. 563-- 569). Cambridge, MA: MIT Press.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC